Resveratrol and para-coumarate serve as ring precursors for coenzyme Q biosynthesis.

Coenzyme Q (Q or ubiquinone) is a redox-active polyisoprenylated benzoquinone lipid essential for electron and proton transport in the mitochondrial respiratory chain. The aromatic ring 4-hydroxybenzoic acid (4HB) is commonly depicted as the sole aromatic ring precursor in Q biosynthesis despite the recent finding that para-aminobenzoic acid (pABA) also serves as a ring precursor in Saccharomyces cerevisiae Q biosynthesis. In this study, we employed aromatic 13C6-ring-labeled compounds including 13C6-4HB, 13C6-pABA, 13C6-resveratrol, and 13C6-coumarate to investigate the role of these small molecules as aromatic ring precursors in Q biosynthesis in Escherichia coli, S. cerevisiae, and human and mouse cells. In contrast to S. cerevisiae, neither E. coli nor the mammalian cells tested were able to form 13C6-Q when cultured in the presence of 13C6-pABA. However, E. coli cells treated with 13C6-pABA generated 13C6-ring-labeled forms of 3-octaprenyl-4-aminobenzoic acid, 2-octaprenyl-aniline, and 3-octaprenyl-2-aminophenol, suggesting UbiA, UbiD, UbiX, and UbiI are capable of using pABA or pABA-derived intermediates as substrates. E. coli, S. cerevisiae, and human and mouse cells cultured in the presence of 13C6-resveratrol or 13C6-coumarate were able to synthesize 13C6-Q. Future evaluation of the physiological and pharmacological responses to dietary polyphenols should consider their metabolism to Q.

Supplementary key words antioxidants • isoprenoids • lipids/chemistry • mass spectrometry • mitochondria • ubiquinone • plant polyphenols • stilbene ( 20 ) and 11 Ubi polypeptides (UbiA-UbiJ and UbiX; Fig.  1 ) ( 21 ). UbiC carries out the fi rst committed step in the biosynthesis of Q 8 , the conversion of chorismate to 4HB ( 22 ). UbiA adds the octaprenyl tail to the 4HB ring, followed by the decarboxylation catalyzed by UbiD and UbiX. UbiI adds the fi rst hydroxyl group at the C5 position, followed by O-methylation catalyzed by UbiG, the homolog of yeast Coq3. Additional ring modifi cations catalyzed by UbiH, UbiE, UbiF, and UbiG generate the fi nal product of Q 8 . UbiB, an atypical protein kinase similar animal cells (15)(16)(17). Yeast can also use para -aminobenzoic acid (pABA) as an alternate ring precursor in the biosynthesis of Q ( 9,18 ). This fi nding was surprising because pABA is a well-known precursor of folate, which is synthesized de novo by many microorganisms and folate is a vitamin for humans. A biosynthetic scheme was reported recently including proposed steps for the conversion of pABA to Q 6 in S. cerevisiae ( 19 ).
The biosynthesis of Q 8 in E. coli requires IspB (which synthesizes the octaprenyl diphosphate tail precursor) Fig. 1. Schemes of Q biosynthesis in S. cerevisiae, other eukaryotes, and E. coli. In S. cerevisiae , Coq1 synthesizes the hexaprenyl diphosphate tail, and Coq2 adds the hexaprenyl tail (denoted as "R") to either 4HB or to pABA, forming 3-hexaprenyl-4HB (HHB) or 3-hexaprenyl 4-aminobenzoic acid (HAB). Coq6 adds the fi rst hydroxyl group to the C5 position of the aromatic ring, forming either 3-hexaprenyl-4,5-dihydroxybenzoic acid (DHHB) or 3-hexaprenyl-5-hydroxy-4-aminobenzoic acid (HHAB). An undetermined enzyme catalyzes the decarboxylation step, forming demethyl-demethoxy Q 6 (DDMQ 6 ) or imino-demethyl-demethoxy-Q 6 (IDDMQ 6 ). Coq5 catalyzes the C -methylation at the C2 position of the aromatic ring, producing either demethoxy-Q 6 (DMQ 6 ) or imino-demethoxy-Q 6 (IDMQ 6 ). The 4HB and pABA branches are proposed to converge at the steps designated by the dotted arrows. Coq7 adds the second OH group to the C6 position, generating demethyl-Q 6 (DmeQ 6 ), followed by the second O -methylation catalyzed by Coq3 to synthesize Q 6 . Coq4, Coq9, Coq10, and Coq11 are required for effi cient Q 6 biosynthesis, but their function is yet to be determined. Human and mouse cells (depicted as "Euk") produce Q 10 and Q 9 via steps similar to those shown for S. cerevisiae . E. coli proteins responsible for Q 8 biosynthesis are designated with green text. UbiC converts chorismate to 4HB. IspB synthesizes the octaprenyl diphosphate tail and UbiA adds the octaprenyl tail (denoted as "R") to the 4HB or pABA to form 3-octaprenyl-4HB (OHB) or OAB. UbiD and/or UbiX catalyze the decarboxylation of the aromatic ring forming OP or OA. UbiI adds the fi rst hydroxyl to the ring to form octaprenylcatechol (OC) or 2-amino-3-octaprenylphenol (OAP ). The pABA branch of the pathway stops at this step, while UbiG O-methylates OC to form 2-methoxy-OP. Additional ring modifi cations catalyzed by UbiH, UbiE, UbiF, and UbiG form the fi nal product Q 8 H 2 . UbiB and UbiJ are required for Q 8 biosynthesis, but their function is yet to be determined. Boxed compounds designate the aromatic ring precursors tested in this study. liquid medium. Following overnight incubation with shaking (250 rpm) at 30°C, yeast cells were transferred into fresh drop out dextrose (DoD) medium ( 18 ). DoD medium contained 2% dextrose, 6.8 g/l Bio101 yeast nitrogen base minus pABA minus folate with ammonium sulfate (MP Biomedicals), and 5.83 mM sodium monophosphate (pH adjusted to 6.0 with NaOH). Amino acids and nucleotides were added as described previously ( 18 13 C 6 -labeled aromatic ring precursors were added to fresh DoD medium and incubated with yeast cells (100 A 600 ) at 30°C for 4 h . Cells were collected by centrifugation and pellets were stored at Ϫ 20°C. The wet weight of each cell pellet was determined by subtracting the weight of the tube from the total weight. Protein assays (BCA assay, Thermo) were performed on yeast cell lysates ( 25 ). For 13 C 6 -coumarate labeling, BY4741 yeast cells were incubated in 5 ml of SD-complete medium at a starting cell density of 0.1 A 600 and incubated at 30°C for 24 h. The yeast cell density after incubation was approximately 6 A 600 .

C 6 ]
The synthesis was similar to the method described by Robbins and Schmidt ( 26 ), with the following modifi cations. To a fl amedried fl ask (25 ml) was added 4-hydroxybenzaldehyde [aromatic-13 C 6 ] (50 mg), malonic acid (75 mg), piperidine (5 l), and pyridine (1 ml). The reaction mixture was stirred under argon at 92°C. The reaction was monitored through thin layer chromatography on 0.25 mm SiliCycle silica gel plates and visualized under UV light and with permanganate or 2,4-dinitrophenylhydrazine staining. Upon completion (12 h), the mixture was sequentially added to 10 ml water, neutralized to pH 7-8, and then washed with dichloromethane. The aqueous solution was acidifi ed to pH 1 and then extracted twice using ethyl acetate. The combined organic extract was concentrated in vacuo and purifi ed through fl ash column chromatography. Flash column chromatography was performed with SiliCycle Silica-P Flash silica gel (60 Å pore size, 40-63 m) and 50% ethyl acetate in hexanes as mobile phase, to furnish an off-white solid (58 mg, 87% yield). A portion was further purifi ed by semi-preparative RP-HPLC (Waters Sunfi re C18, to Coq8, and UbiJ play essential, but unknown, functions in E. coli Q 8 biosynthesis ( 21 ).
Recently, Block et al. ( 15 ) identifi ed para-coumarate (pcoumarate) as a ring precursor of Q biosynthesis in Arabidopsis thaliana . Arabidopsis converts phenylalanine to p-coumarate in the cytosol, and following transport into peroxisome, p-coumarate is ligated to CoA and the threecarbon side chain is shortened via peroxisomal ␤ -oxidation ( 15 ). Plant peroxisomes appear to contain thiolases and CoA thioesterases that can ultimately produce 4HB from 4-hydroxybenzoyl-CoA ( 15 ). Tyrosine can also supply the ring of Q in Arabidopsis ; but this must occur via a nonintersecting pathway, because Arabidopsis mutants unable to utilize phenylalanine still utilized tyrosine as a ring precursor of Q ( 15 ). Animal cells are able to hydroxylate phenylalanine to form tyrosine, and it is presumed that conversion of tyrosine to 4HB occurs via its metabolism to p-coumarate ( 16,23 ). However, the enzymes involved in 4HB biosynthesis in either yeast or animal cells have not been identifi ed.
The in vivo metabolism of potential ring precursors labeled with the stable isotope 13 C can be determined with high sensitivity and specifi city with reverse phase (RP)-HPLC-MS/MS identifi cation and quantifi cation. Using this approach, Block et al. ( 15 ) showed that Arabidopsis was not able to incorporate 13 C 6 -pABA into Q. Here, we have made use of 13 C 6 -ring-labeled forms of pABA and p-coumarate to track their metabolic fate as potential Q biosynthetic precursors in E. coli , S. cerevisiae, and animal cells. Due to its structural similarity with p-coumarate, 13 C 6 -resveratrol was also tested as a ring precursor in Q biosynthesis. In this study, we found that human and E. coli cells do not utilize pABA as an aromatic ring precursor in the synthesis of Q, while resveratrol and p-coumarate serve as ring precursors of Q in E. coli, S. cerevisiae, and human cells.

Yeast growth and stable isotope labeling
The S. cerevisiae strains used are described in Table 1 . YPD medium (2% glucose, 1% yeast extract, 2% peptone) was prepared as described ( 24 ). Solid plate medium included the stated components plus 2% Bacto agar. Yeast colonies from YPD plate medium were fi rst inoculated into 250 ml fl asks containing 70 ml YPD  , and released from the culture dish with 0.25% trypsin-EDTA (Gibco). Aliquots of the released cells were stained with Trypan blue and the number of cells counted with the Cellometer Auto T4 (Nexcelom Bioscience); aliquots (5%) were also removed for determination of protein content (BCA assay; Thermo). The remaining cells in the suspension were collected by centrifugation. Cell pellets were stored at Ϫ 20°C.
NMR spectra were recorded using a Bruker Avance-500 spectrometer, calibrated to residual acetone-d 6 as the internal reference (2.05 ppm for 1 H NMR; 29.9 and 206.7 ppm for 13 C NMR). 1 H NMR spectral data are reported in terms of chemical shift ( ␦ , parts per million), multiplicity, coupling constant (hertz), and integration . 13 C NMR spectral data are reported in terms of chemical shift ( ␦ , parts per million), multiplicity, and coupling constant (hertz). The following abbreviations indicate the multiplicities: s, singlet; d, doublet; t, triplet; m, multiplet; br, broad. 1

E. coli growth and stable isotope labeling
E. coli strains are described in Table 1 . The BW25113 ⌬ ubiC::kan mutant strain was obtained from the Keio collection ( 27 ). Phage P1 was used to transduce the mutation into the MG1655 strain, yielding MG1655 ubiC . The replacement of the chromosomal ubiC gene by the kan gene was checked by PCR amplifi cation. Cells were inoculated in 100 ml of Luria broth (LB) for 16 h at 37°C. Cells (50 A 600 ) from each sample were collected by centrifugation, and the collected pellets were resuspended in fresh LB medium in the presence of either 10 g/ml of 13 C 6 -4HB, 13 C 6 -pABA, or 13 C 6 -resveratrol, and incubated at 37°C with shaking at 250 rpm. Incubations with vehicle control contained an equivalent volume of ethanol (in all conditions the fi nal ethanol concentration was 0.2%). Cells were collected by centrifugation after 4 h and stored at Ϫ 20°C for LC-MS/MS lipid analyses. For 13 C 6coumarate labeling, HW272, HW25113, MG1655, and MG1655u-biC were inoculated in 5 ml of LB for 16 h at 37°C (MG1655ubiC was incubated in LB with 50 g/ml kanamycin). Cells were diluted to 0.2 A 600 in fresh media with 15 g/ml of 13 C 6 -coumarate and incubated for 24 h. Cells were pelleted for lipid extraction and LC-MS/MS analyses.
Animal cell culture and stable isotope labeling U251 human glioma and 3T3 mouse fi broblast cells were cultured in DMEM (Gibco). U87 human glioma cells were cultured in Iscove's Modifi ed Dulbecco's Medium (Gibco). Human embryonic kidney 293T cells were cultured in DMEM with 1 mM sodium pyruvate (Gibco). All cells were passaged in the stated media supplemented with 10% FBS (Omega Scientifi c) and 1% penicillin-streptomycin (10,000 U/ml) (Life Technologies). Equal numbers of cells were plated approximately 12 h prior to treatment experiments. During treatment with stable Luna phenyl-hexyl column (particle size 3 m, 50 × 2.00 mm; Phenomenex) for mammalian and bacteria cell lipid extracts. The mobile phase consisted of solvent A (methanol:isopropanol, 95:5, with 2.5 mM ammonium formate) and solvent B (isopropanol, 2.5 mM ammonium formate). For separation of yeast quinones, the percentage of solvent B increased linearly from 0 to 5% over 6 min, and the fl ow rate increased from 600 to 800 l/ min. The fl ow rate and mobile phase were linearly changed back to initial condition by 7 min. For separation of bacteria and mammalian quinones, the percentage of solvent B for the fi rst 1.5 min was 0%, and increased linearly to 10% by 2 min. The percentage of solvent B remained unchanged for the next min and decreased linearly back to 0% by 6 min. A constant fl ow rate of 800 l/min was used. All samples were analyzed in multiple reaction monitoring mode; multiple reaction monitoring transitions were as follows: yeast lipid extracts. Diethoxy-Q 10 ( 28 ) was used as an internal standard for determination of Q 9 and Q 10 in mammalian cell lipid extracts and Q 8 in E. coli cell lipid extracts. Samples were vortexed for 45 s, then the upper layer was removed to a new tube, and another 1.8 ml of petroleum ether was added to the lower phase and the sample was vortexed again for 45 s. The upper layer was again removed and combined with the previous organic phase. The combined organic phase was dried under a stream of nitrogen gas and resuspended in 200 l of ethanol (USP; Aaper Alcohol and Chemical Co., Shelbyville, KY).

RP-HPLC-MS/MS
The RP-HPLC-MS/MS analyses were performed as previously described for determination of Q 6 in yeast lipid extracts ( 11,18 ) and determination of Q 9 and Q 10 in mammalian lipid extracts ( 28,29 ). Briefl y, a 4000 QTRAP linear MS/MS spectrometer from Applied Biosystems (Foster City, CA) was used. Applied Biosystem software, Analyst version 1.4.2, was used for data acquisition and processing. A binary HPLC solvent delivery system was used with either a Luna phenyl-hexyl column (particle size 5 m, 100 × 4.60 mm; Phenomenex) for yeast cell lipid extracts or a independent of the supplied 13 C 6 -pABA, suggesting that OAP might be a "dead-end" product. 13 C 6 -OAB, 13 C 6 -OA, or 13 C 6 -OAP were not detected in either the 13 C 6 -4HB-treated or control E. coli cells ( Fig. 3A, B, D ). 13 C 6 -OP was detected only in 13 C 6 -4HB-treated cells ( Fig. 3C ). These results suggest that although pABA is prenylated and can be further modifi ed by UbiD, UbiX, and UbiI, E. coli may not be able to process the aniline-containing ring intermediates to later intermediates or to Q 8 .
Given that S. cerevisiae can utilize either 4HB or pABA in Q 6 biosynthesis, we investigated the use of other possible aromatic ring precursors. Surprisingly, wild-type yeast could use resveratrol as a ring precursor in the synthesis of Q 6 ( Fig. 4 ). W303 cells cultured in the presence of 68, 282, or 974 M of 13 C 6 -resveratrol showed increasing amounts of 13 C 6 -Q 6 , while the ethanol control samples contained no detectable 13 C 6 -Q 6 . Notably, the increased amount of resveratrol did not alter the total Q 6 content. We next examined whether human or mouse cells could use resveratrol as a ring precursor to Q. The three human cell lines we examined were able to convert resveratrol to Q, as shown by the accumulation of 13 C 6 -Q 10 ( Fig.  5A-C , supplementary Fig. 3A-C). 13 C 6 -Q 9 ( Fig. 5D , supplementary Fig. 3D) also accumulated in mouse 3T3 fibroblasts, when cultured in the presence of 70 M of 13 C 6 -resveratrol. Although cells cultured with 13 C 6 -4HB accumulated signifi cantly more 13 C 6 -Q 10 than when cultured with 13 C 6 -resveratrol, the incorporation of 13 C 6resveratrol into 13 C 6 -Q 10 accounted for approximately 10% of the total Q 10 , a proportion that was much higher than that observed in wild-type yeast cells (the 13 C 6 -Q 6 was less than 1% of the total Q 6 ). 13 C 6 -Q 10 content in U251 and 13 C 6 -Q 9 3T3 cells increased in response to the

RESULTS
pABA is a demonstrated ring precursor of Q biosynthesis in the yeast S. cerevisiae ( 9,18 ), but is not utilized as a ring precursor of Q biosynthesis in Arabidopsis ( 15 ). To investigate whether pABA may serve as a ring precursor of Q biosynthesis in mammalian cells, human U251 cells were cultured in the presence of 7, 70, or 700 M of either 13 C 6 -4HB or 13 C 6 -pABA for 24 h prior to RP-HPLC-MS/MS analysis of Q content ( Fig. 2A ). U251 cells readily converted 13 C 6 -4HB to 13 C 6 -Q 10 , however, incubations with 13 C 6 -pABA produced no detectable 13 C 6 -Q 10 ( Fig. 2A , supplementary Fig. 2). Treatments of U251 cells with various 13 C 6 -4HB concentrations did not alter the total Q 10 content; however incubation with 700 M 13 C 6 -pABA resulted in signifi cantly lower total Q 10 content in mammalian cells ( P < 0.05).
To examine whether pABA is utilized as a ring precursor in E. coli Q 8 biosynthesis, 13 C 6 -pABA or 13 C 6 -4HB was added to cultures of the designated E. coli strains. HW272 and MG1655 are wild-type strains, while MG1655 ubiC contains a deletion of the ubiC gene encoding chorismate pyruvate lyase ( Table 1 ). Each E. coli strain was cultured in LB medium with aromatic ring precursors added to a fi nal concentration of 10 g/ml ( Fig. 2B ). Each of the E. coli strains incubated in the presence of 10 g/ml 13 C 6 -4HB accumulated signifi cant amounts of 13 C 6 -Q 8 . No incorporation of 13 C 6 -pABA into 13 C 6 -Q 8 was detected with the wild-type strains. Interestingly, the E. coli ubiC mutant was also unable to use pABA to synthesize Q 8 . This result suggests that pABA is not utilized, even under conditions of impaired 4HB synthesis.
Detection of various polyprenylated derivatives of 13 C 6 -pABA indicated that the E. coli strains tested were able to take up this ring. For example, 13 C 6 -3-octaprenyl-4-aminobenzoic acid (OAB) indicated that 13 C 6 -pABA-treated E. coli cells successfully absorbed 13 C 6 -pABA from the medium and performed the ring prenyltransferase step catalyzed by UbiA ( 30 ) ( Fig. 3A ). 13 C 6 -OA ( Fig. 3B ) was also readily detected in lipid extracts of the 13 C 6 -pABA-treated E. coli cells. Notably, the ubiC mutant accumulated signifi cantly more anilinecontaining intermediates ( Fig. 3A, B ), even in the absence of pABA addition, presumably due to a defi ciency in 4HB synthesis. The product 13 C 6 -OAP ( Fig. 3D ) was also observed in 13 C 6 -pABA-treated E. coli cells and is probably due to UbiI, which catalyzes the fi rst hydroxylation step in Q 8 biosynthesis ( 31 ). The ubiC mutant accumulated 10 times more 12 C-OAP than HW272 or MG1655, a fi nding   4B). 13 C 6 -OP was detected only when E. coli strains were cultured in the presence of 13 C 6 -4HB, and not with 13 C 6 -resveratrol ( Fig. 6B ), suggesting that step(s) at which resveratrol is used as a ring precursor may not depend on its conversion to 4HB, or that the production of 4HB from resveratrol is slow compared with the step where OP is utilized.
Given the structural similarity of resveratrol with p-coumarate, we tested the ability of yeast to utilize 13 C 6 -coumarate as a ring precursor of 13 C 6 -Q 6. Yeast wildtype BY4741 cells were cultured in SD-complete medium with 7, 70, or 700 M of either 13 C 6 -4HB or 13 C 6 -coumarate for 24 h ( Fig. 7A ). We found that while the total amount of Q 6 did not change with different amounts of 13 C 6 -coumarate, the amount of 13 C 6 -Q 6 increased with higher concentrations of 13 C 6 -coumarate, although the incorporation was lower as compared with 13 C 6 -4HB. U251 human cells were labeled increasing concentrations of 13 C 6 -resveratrol, while the total Q content again remained unaltered ( Fig. 5E, F ). Unfortunately, higher concentration (>70 M) of resveratrol induced cell death, thus we were not able to examine the amount of 13 C 6 -Q synthesized in the presence of higher 13 C 6 -resveratrol concentrations.
E. coli also utilized resveratrol as an alternative ring precursor to Q, although to a lesser extent when compared with yeast, mouse, or human cells. HW272 and MG1655 cells cultured in LB medium in the presence of 10 g/ml 13 C 6 -resveratrol accumulated trace amounts of 13 C 6 -Q 8 ( Fig. 6A , supplementary Fig. 4A). In comparison, 10 g/ml of 13 C 6 -4HB resulted in 13 C 6 -labeling of more than two-thirds of the total Q content in the same cells. However, the E. coli ubiC mutant, with a defect in de novo synthesis of 4HB, produced signifi cantly more 13 C 6 -Q 8 when treated with 13 C 6 -resveratrol ( Fig. 6A , supplementary S. cerevisiae cells could utilize pABA as a ring precursor in Q biosynthesis was rather surprising because pABA is a crucial intermediate in folate biosynthesis ( 9,18 ). The addition of pABA to either E. coli or human cells leads to a concentration-dependent inhibition of Q content ( 4,32,33 ). Another aromatic ring compound, 4-nitrobenzoic acid, inhibited Q biosynthesis in mammalian cells by competing with 4HB for Coq2 ( 34 ). While pABA does not function as a ring precursor of Q in Arabidopsis ( 15 ), it remained possible that pABA might still be utilized as a ring precursor in Q biosynthesis in human and E. coli cells. Therefore, we employed 13 C 6 -pABA to investigate its fate in human and E. coli cells.
Treatment of cells with 13 C 6 -pABA revealed that pABA was not an aromatic ring precursor to Q biosynthesis in either human or E. coli cells. In order to rule out the scenario that E. coli cells might utilize pABA as a ring precursor in Q biosynthesis only when the primary ring precursor 4HB is not available, we incubated ubiC mutants, which have defects in the de novo synthesis of 4HB in the presence of 13 C 6 -pABA. However, even ubiC mutants were not able to utilize pABA for Q 8 biosynthesis. Interestingly, we detected multiple nitrogen-containing intermediates that derived from 13 C 6 -pABA. Detection of 13 C 6 -OAB in all three strains (HW272, MG1655, and ubiC ) confi rmed 13 C 6 -pABA uptake ( Fig. 3A ). Further modifi cations of the 13 C 6 -OAB resulted in 13 C 6 -OA and 13 C 6 -OAP, indicating UbiA, UbiD/UbiX, and UbiI tolerated the amino ring substituent ( Fig. 3B, D ) ( 21 ). OAP accumulated in the ubiC mutant independent of 13 C 6 -pABA addition, suggesting that OAP could be a dead-end product derived from endogenously produced unlabeled pABA. Neither HW272 nor MG1655 wild-type E. coli accumulated signifi cant amounts of OAP, indicating that E. coli cells tend to process pABA through early steps in the Q biosynthetic pathway when 4HB content is low. These observations are consistent with studies that showed an E. coli mutant that lacked chorismate synthase converted pABA to OAB when cultured without addition of 4HB ( 33 ). We did not detect further downstream nitrogencontaining Q biosynthetic intermediates using targeted and limited-untargeted LC-MS/MS approaches. However, the presence of additional N-containing Q intermediates downstream of OAP cannot be ruled out.
It was shown that Lithospermum erythrorhizon cell cultures are able to synthesize 4HB from p -coumarate ( 35 ) and A. thaliana uses p -coumarate to synthesize Q ( 15 ). Therefore, we investigated whether p -coumarate is a ring precursor for Q in different organisms. We found that yeast, E. coli , and human cells can derive Q from p -coumarate. This fi nding will help us understand how 4HB is generated in these organisms. In A. thaliana , p -coumarate is activated by CoA ligase and the aliphatic chain is shortened to 4HB in peroxisomes ( 15 ). Because yeast, human cell cultures, and E. coli can use p-coumarate to make Q, it is possible that these organisms derive 4HB from p-coumarate in a similar manner.
A wide spectrum of activities is attributed to stilbenoids produced by a variety of plants when under attack by pathogens ( 36 ). A stilbenoid of recent fame, resveratrol, with 7, 70, or 700 M of either 13 C 6 -4HB or 13 C 6 -coumarate for 24 h ( Fig. 7B ). We found that more 13 C 6 -Q 10 accumulated when U251 cells were treated with increasing concentrations of 13 C 6 -coumarate. Similar to yeast cells, U251 cells showed enhanced conversion of 13 C 6 -4HB to Q as compared with 13 C 6 -coumarate. Finally, we investigated the conversion of p-coumarate to Q 8 in E. coli . The designated wild-type E. coli strains and the ubiC mutant were labeled with 15 g/ml 13 C 6 -coumarate for 24 h. 13 C 6 -coumarate was converted to 13 C 6 -Q 8 much more effi ciently in ubiC mutants than in the wild-type E. coli strains ( Fig. 7C ). The results show that pcoumarate is a ring precursor for Q biosynthesis in S. cerevisiae , E. coli , and human cells.

DISCUSSION
Most schemes of Q biosynthesis continue to depict 4HB as the "sole" aromatic ring precursor. The fi nding that effects of resveratrol have led to vigorous research investigating its mechanisms of action.
Here we show that resveratrol serves as an aromatic ring precursor in Q biosynthesis in E. coli , yeast, and mammalian cells. Wild-type E. coli barely utilized resveratrol for Q biosynthesis; however, signifi cant incorporation of the resveratrol ring into Q 8 was observed in ubiC mutants. Preferential incorporation of alternate ring precursors in the E. coli ubiC mutant strain is presumably due to the defect in synthesis of 4HB. In contrast, approximately 10% of the total Q content in human and mouse acts as a chain-breaking antioxidant, modulates cellular antioxidant enzymes and apoptosis, and has benefi cial effects on neurodegenerative and cardiovascular diseases, eliciting metabolic responses similar to dietary restriction ( 37,38 ). Although there is much controversy regarding the lifespan extension effects of resveratrol ( 39 ), its effects on age-associated diseases in animal models has generated considerable enthusiasm for research on its mechanism of action ( 40 ). Many questions remain regarding resveratrol biodistribution, its metabolism, and the biological effects of resveratrol metabolites ( 41 ). The benefi cial health cells harbored the ring derived from 13 C 6 -resveratrol after 24 h of incubation. The maximum concentration of resveratrol tested was lower than either 4HB or pABA because resveratrol induced apoptotic cell death ( 42 ). Thus, we monitored cell viability in our experiments and limited the amount of resveratrol tested in order to avoid induction of cell death.
The metabolism of resveratrol responsible for its incorporation into Q has not been determined. Animals harbor two carotenoid cleaving enzymes, BCO1 and BCO2, and both are homologs of the carotenoid cleavage oxidase family ( 43 ). BCO1 is cytosolic and is responsible for cleaving ␤ -carotene to form two molecules of retinal, while BCO2 is located in the inner mitochondrial membrane and acts on xanthophylls ( 44 ). It is tempting to speculate that BCO2, which has broader substrate specifi city, might possibly cleave stilbenoids to produce two ring aldehyde products. Other family members of carotenoid cleavage enzyme in bacteria and fungi cleave resveratrol to produce 4-hydroxy-benzaldehyde and 3,5-dihydroxy-benzaldehyde ( 45,46 ). Notably the 4 ′ hydroxyl group of resveratrol has been identifi ed as crucial for antioxidant and neuroprotective effects of stilbenoids ( 47 ). It seems likely that other stilbenoids may serve as ring precursors of Q. For example, processing of piceatannol ( trans-3,5,3 ′ 4 ′tetrahydroxystilbene) by a fungal carotenoid cleavage oxidase family member ( 48 ), generates 3,4-dihydroxybenzaldehyde, a ring precursor that could potentially bypass the Coq6 hydroxylase step of Q biosynthesis upon Coq2-prenylation ( 49 ).
Of the more than one hundred clinical trials testing the effi cacy of resveratrol or other polyphenols (clinical trials.gov), few determine the metabolic fate of the administered supplement. When metabolism of resveratrol is studied, the focus is on aqueous soluble polar metabolites of resveratrol, including sulfated and glucuronidated conjugation products ( 50 ). The new fi nding that a metabolic conversion of resveratrol into Q occurs in eukaryotes shows that exogenous antioxidants may be utilized as precursors to synthesize a wholly different class of molecule. The effects of resveratrol in mimicking calorie restriction ( 37, 38 ) may be due in part to its conversion to Q, a lipid known to induce anti-infl ammatory responses ( 51 ), an essential component of mitochondrial energy metabolism, and a potent lipid soluble antioxidant ( 4 ). Investigation of the pharmacological responses to diverse dietary polyphenols (e.g., curcumin) should be expanded to include this molecular fate. Further investigation on this subject will give us a better understanding on the origin of the benzenoid moiety of Q in different organisms.
The authors thank Drs. Laurent Loiseau, Frederic Barras, and Fabien Pierrel for the generous gift of the MG1655 and MG1655 ubiC E. coli strains. They also thank F. Pierrel for important suggestions regarding the [ 13 C 6 -pABA] labeling studies. They acknowledge the University of California at Los Angeles Molecular Instrumentation Core proteomics facility for the use of QTRAP 4000, National Institutes of Health